The HowardJournal Vol27 No 4. Nou 88 ISSN 0265552’7

An Assessment of the Design, Implementation and Effectiveness of

Neighbourhood Watch in London

TREVOR BENNETT Senior Research Associate, Institute of Criminology,

University of Cambridge

Abstract: This, paper presents the findings of a recently completed evaluation o f two Neighbourhood Watch schemes in London. The main research method was based on crime and public attitude surveys of the two Neighbourhood Watch areas before the schemes were implemented and again oneyear after their implementation. Similar surveys were conducted in a displacement and a control area as a comparison. The results showed that crime increased slightly in the scheme areas and remained constant or fe l l in the displacement and control areas. More favourable results were shown in changes in public attitudes and behaviour. The paper concludes that the lack o f a more substantial success was most likely due to implementation

failure and the weak design of Neighbourhood Watch in the Metropolitan Police District (MPD).

The launch of Neighbourhood Watch (NW) in London on a force-wide basis on 6 September 1983 was in many ways a remarkable event. Neighbourhood Watch was a relatively new concept. The earliest versions of NW appeared in the United States only a decade earlier (Washnis 1976; Cirel et al. 1977) and NW in Britain had a history of less than two years (Anderton 1985). It was also remarkable in the scale of the operation. The Commissioner’s plan for every division in the MPD to be covered at least in part under the umbrella of NW represented a crime prevention initiative of grand proportions. Perhaps the most remarkable aspect of the launch, however, was the commitment to a major project which neither had been fully piluted nor evaluated. The launch of NW in London was not only a courageous and spirited act on the part of the Commissioner, but also a highly speculative one.

One of the aims of this paper is to present some of the findings of a recent evaluation of two NW schemes in the MPD and to discuss the extent to which the Commissioner’s speculation has produced returns. In addition, the paper aims to examine the theory and practice of NW and to offer some proposals on possible future developments of NW in London. I t is not the aim of the paper to present complicated statistical findings nor to engage in lengthy justifications of the research method. Such discussions can be found elsewhere (Bennett 1987; Bennett forthcoming). In order to present the main results as accurately as possible, however, it

241

is necessary to make some reference to the methods and statistical analysis used.

Before examining the research evidence it is helpful to look more closely a t the nature of the programme being investigated. There is no common agreement on what precisely NW is or even what precisely it should be called. Schemes around the country vary in terms of the activities they perform and in terms of the labels under which they operate (for examplc, Community Watch, Home Watch, Apartment Watch and Block Watch). O n e reason for this variation is differences in the way in which the concept of NW is interpreted and operationalised. In order to discover the way in which NW in the MPD has been operationalised, it is necessary to look a t the published policy documents that emerged about the time of its inception.

The earliest official documents which serve to define NW in the M P D were principally the ‘A’ Department Memorandum and the Force Instructions (Assistant Commissioner ‘A’ Department 1983a, 1983b), and the official guidelines (Russell undated). ‘The concept of NW as outlined in these papers hinges on the notion of the public becoming the ‘eyes and ears of the police’. ’The Force Instructions (Assistant Commissioner ‘A’ Department 1983a) elaborate this idea more fully:

Neighbourhood Watch is primarily a network of public spirited members of the community, who observe what is going on in their own neighbourhood and report suspicious activity to the police. In simple terms the citizen becomes the ‘eyes and ears’ of the police . . . (p. 1)

Another important feature of NW in London is that i t is conceived as a comprehensive package. The four elements of the package are outlined in the Force Instructions as (i) Neighbourhood Watch; (ii) Property Marking; (iii) Home Security Surveys; and (iv) Environmental Aware- ness. The latter ofthese elements is described in more detail in the official guidelines as an attempt:

to encourage members to consider their surroundings and put forward suggestions regarding alterations and improvements that could remove the opportunity for criminal nuisance . . . (Russell undated, p. 2)

In order to assess the effectiveness of NW, it is first necessary to identify its aims and what it hopes to achieve. I t is clear from these documents that the primary aim of NW as conceived by the Metropolitan Police is to reduce crime. This objective is elaborated more fully in the ‘A’ Department Memorandum ( 1983b) which spotlights ‘opportunist crime’ and residential burglary as the primary targets. Other subsidiary crimes are identified in the Force Instructions including: street robbery, vehicle crime and criminal damage as offences which NW might be able to do something about. A second important aim of NW in London is the reduction of fear of crime, as outlined in the Force Instructions; improvements in public awareness to safeguard property; greater contact between neighbours; and closer liaison between the police and the public, as outlined in the official guidelines.

24 2

It is also important in evaluating social experiments that the mechan- isms by which the programme hopes to achieve its aims are made clear. The main process identified in the policy documents by which NW might reduce crime is through ‘opportunity reduction’. The key mechanism in this process is the greater willingness of members of the public to look out for suspicious incidents and report them to the police. The way in which this might reduce crime is necessarily speculative. I t is possible that the reporting of suspicious incidents by the public to the police improves arrest and conviction rates, which depletes the pool of offenders operating in the NW area. I t is also possible that potential offenders become aware that members of the public in NW areas actively look out for suspicious behaviour and are deterred from offending in those areas.

The remainder of this paper will summarise the main findings of an evaluation of two NW schemes in the MPD to determine the extent to which these aims were achieved. The broad parameters of the research were discussed in collaboration with the Home Ofice (who funded the research) and the Metropolitan Police (who provided access to conduct i t) . As a result of these discussions the original terms of reference specified that the evaluation should focus on one or two NW schemes. At the time, the Metropolitan Police were conducting their own ‘global’ evaluation of NW using police recorded crime as the data source. The police were aware of the limitations of using police recorded crimes and the problems associated with changes in public reporting and police recording practices. It was hoped that the independent evaluation would comple- ment this research by using alternative data sources. The original terms of reference also specified that the study should investigate the impact of NW both on crime and on the general quality of life of residents living in the NW areas.

Method The research method was based on an evaluative approach used frequently by social psychologists known as a quasi-experimental design (Cook and Campbell 1979; Judd and Kenny 1981). The main feature of this design is the collection of data (for example, questionnaire responses from individuals or groups) at a point before the implementation of some kind of treatment (in this case NW) and to repeat the operation at a later point after the treatment has been given time to take effect. This is often accompanied by the collection of data from control individuals or groups who are not receiving the treatment (for example, non-NW participants or areas).

The broad plan of the NW evaluation was to draw upon the principles of this approach and to identify two areas which were about to implement NW schemes (experimental areas) and to compare these with one non- NW area adjacent to one of the scheme areas (a ‘displacement area’) and one non-NW area some distance from either of the scheme areas (a ‘control’ area).

The choice of experimental areas (areas about to receive NW) was

243

guided by a number of factors. First, it was decided that the schemes should be of a kind that had good chance of success from the outset. I f the most promising schemes were later shown to have little effect on crime, it would not be expected that less promising schemes would be any more successful. Second, the schemes had to cover a fairly large number of households in order to generate a suficient number of crimes to be able to conduct a meaningful statistical analysis. Third, it was important that the schemes were likely to be well implemented and sustained for the experimental one-year period.

The experimental areas were selected with the assistance of the Metropolitan Police Crime Prevention Branch. The first stage of the srarch involved identifying divisions which in the past had implemented large schemes and which had a record of obtaining above-average participation rates. In the second stage senior officers from these divisions were contacted to establish whether they had plans to implement similar schemes in the near future. ‘The third stage of the search required visits to a small number of areas to assess their suitability. The selection criteria for the experimental area included the following: at least 500 households, an average or above-average crime rate, a fairly consistent crime rate over the previous two years, actual or symbolic boundaries, enthusiastic potential area co-ordinators, enthusiastic senior oficers, enthusiastic crime prevention and home beat officers. Two sites were eventually selected: one in Acton and another in Wimbledon.

The single displacement site was chosen from areas adjacent to the Wimbledon experimental area. The local police could not guarantee that any of the areas adjacent to the Acton site would remain NW free for the next twelve months. The important selection criteria were that it should adjoin the experimental area along at least part of its boundary and that there should be no plans to implement NW in the area in the foreseeable future. In addition, an attempt was made to match the displacement area to the experimental area in terms of social composition, general grographic structure and crime rate.

The control site was chosen to match the Wimbledon experimental site as this had a suitable displacement area. The selection was made by matching the Wimbledon area with a randomly chosen area of similar social composition and of similar distance from the centre of London. The social composition of the experimental and control areas was matched using .\CORN typing which identifies enumeration district in terms of census variables (see Hough and Mayhew 1985 for details of the use of ACORN typing in the British Crime Survey). As the Acton and Wimbledon areas were similar in social composition (two of the three E.D.’s in each area were of the same ACORN classification), it was felt that the single control site could be used for both.

l h e main method of data collection used in the research was a crime and public attitudes survey of the four areas (two experimental, one displacement and one control) before the launch of the NW schemes and again one year following the launch. Because of the small number of crimes that might be committed in areas of this size, it was decided to use

244

a 100% sample of the area and to attempt to interview a randomly selected adult representative of every household.

The questionnaire was a modified version of the schedule used in the first and second British Crime Surveys (Hough and Mayhew 1983, 1985). It was divided into five main parts. The first part was used to record details of contacts made and not made. The second part covered fear of crime, residents’ satisfaction with their area, perceived probability of victimisation and screening questions relating to victimisations over the previous twelve months (used to determine which victimisations should be recorded in more detail in the third part of the questionnaire). The third part was used to record details of offences identified in the screening questions. The fourth part included questions about NW and about the police. The fifth part included questions about the social and demo- graphic characteristics of the respondent and the respondent’s household.

In addition to the crime surveys, telephone calls made by residents in the NW areas to the police including station calls and emergency (‘999’) calls were recorded in order to monitor the impact of the programme on the flow of information and reports of suspicious activities.’

The Samples

The first round of surveys was conducted in May 1985, between one and two months before the launch of the NW programmes in the two experimental areas. The second round of surveys took place in July 1986 after the programmes had been running for about one year. Details of the number of interviews achieved in both rounds of surveys are shown in Table I.

The overall contact rate was 62% in the pretest period (‘before’) surveys and 64% in the post-test period (‘after’ surveys). The main reasons for non-contacts was ‘no contact after four or more call backs’ and ‘personal refusal’.

The Programmes

The design of the NW programmes in the experimental areas followed the broad framework of NW in the MPD. The programmes were launched with a public meeting held in the proposed NW area, attended by representatives of the local police and about 100 members of the public, The police outlined the principles of NW as they saw them and encouraged members of the public in the area to participate. NW and other crime prevention documents were handed out to members of the audience and potential street co-ordinators were recruited. About three months following the official launch meeting street signs were erected at visible points at every street entrance leading into the NW area.

I t is important to the evaluation to determine the extent to which the experimental programme was implemented in the area. Table 2 provides details of levels of participation among residents in the two experimental areas in relation to ten key elements associated with NW.

245

TABI,E I Response Rates and Sample Sizes f o r the Four Survey Areas

A C T . E X P . ‘ WIMB.EXP. W I M B . D I S P . RED.CON.

Pre- test T o t a l h.h.’s2 639 T o t a l inelig.’ 115 T o t a l elig. 524

Interviewed 306 Response rate4 58%

Post-test T o t a l h.h.’s 702 ‘Total inelig. 179 Total elig. 523

Interviewed 309 Response rate 59%

711 7 3

638

353 55%

710 95

615

400 65 Yo

540 74

466

323 69%

570 87

483

332 69 Yo

495 53

442

306 69%

557 44

513

330 64%

Notes: 1 ACT.EXP = Acton cxpcrimcntal area; WIMB.EXP. = Wimblcdon cxpcrimcntal arca; WIMB.DISP. = Wimblcdon displacement arca; REI).CON. = Rcdbridgc control area. 2 h.h.’s = households 3 Addresses wcrc incligiblc i l thcy wcrc vacant, dcmolishcd or non-residential. Households wcre incligiblc if thc houscholdcrs had not livcd in the arca for at lcast onc year prior to intcrvicw. 4 Rcsponsc rate is the total number of interviews dividcd by the total number of cligiblc households.

Table 2 shows that nearly all respondents in the second round of surveys were aware that a NW scheme was operating in their area. About 62% of respondents in Acton and 44% of respondents in Wimbledon defined themselves as participants in the scheme and about the same percentage in each area said that they had displayed a NW stickerin their windows. I t could be argued that this is a reasonable level of commitment to NW. It is quite exceptional to launch a crime prevention campaign which results in almost half of the target population taking an active part in it. It is also difficult to argue that nothing happened in the areas (in terms of programme implementation) when almost half of all households in one area and nearly two-thirds in another claimed to be supporting it. Just under half of the respondents said that they had deliberately looked out for anything suspicious over the last year. In other respects, however, levels of participation were not so encouraging. Only a small percentage of respondents knew by name or by site their home beat officer. Few respondents had marked any of their property and few had received a home security survey.

246

TABLE 2 Involvement of Residents in the Programme Elements in the Two Experimental Areas

(percentages)

ACTON EXP. WIMBLEDON EXP.

Yes No’ Total Yes No’ Total ( ~ 4 0 0 ) (n=309)

Know NW in area2 Attended launch3 Received newsletter4 Knew co-ord.5 Knew beat off..“ Participant7 Disp. NW sticker’ Disp. PM sticker9 Home sec. survey“’ Looked out’ I

95 5 100 13 77 100 89 1 1 100 61 39 100 28 72 100 62 38 100 64 36 100 8 92 100 5 95 100

47 53 100

90 10 100 17 73 100 46 54 100 21 79 100 15 85 100 44 56 100 40 60 100 24 76 100

3 97 100 40 60 100

Notes: 1 The category ‘No’ includes ‘don’t know’ and ‘missing’ responses. 2 Did you know that a neighbourhood watch scheme was running in this area? 3 Did you or any members of your household attend the launch meeting of the

4 Have you received a newsletter from the neighbourhood watch scheme? 5 Do you know your street co-ordinator by name or by sight? (Coded as ‘yes’ if

6 Do you know your local Home Beat Officer by name or by sight? 7 Would you say that you are a member of, or participant in, the neighbourhood

watch scheme? 8 Have you displayed a neighbourhood watch scheme sticker like this one on one

of your doors or windows? 9 Have you displayed a property marking sticker like this one on one of your

doors or windows? 10 Have you had a visit from a police crime prevention officer or any other officer to advise you on security in your home? 1 1 In the last twelve months, have you deliberately looked out in the area for anything suspicious?

scheme?

respondent is the street co-ordinator).

Results

The main findings relating to the effectiveness of NW concern its impact on victimisation. The results presented here derive from survey respondent reports of victimisations against their household or against them in person over the twelve mouth period before the surveys. The main offences covered by the survey include the usual household and street crimes that are most prevalent in small neighbourhoods (see Table 3 for the main

247

TARTAR 3 Households Victimised Ow or More Times in the Pre-lest and Post-test Periods in the

i;our Survey Areas (percentages)

ACT.EXP.' WIMB.EXP. WIMB.l)ISI'. RED.CON. ~~

Pre Post Pre Post Pre Post Prc Post n = 306 309 353 400 323 332 306 330

Thef t of vehiclc' 1 2 I 1 1 1 1 I Thcft from veh.' 3 4 3 6 6 5 3 2 Crim. dam. to

vrh ' 4 8 6 10 6 7 4 3 Burglary

dwelling' 6 n 5 4 7 0 3 2 Other crim. dam." 2 4 2 5 5 7 :1 2

All h.hold ofTs.' 17 26 20 23 25 29 15 1 I All pers. offs." 1 5 2 2 4 3 1 2

NolP,: 1 AC'I.EXP. = Acton rxpcrimcntal arra; WIMB. EXP. = Wimblcdon experimental area; WIMR. DISP. = Wimblcdon displaccmcnt arca; RED. CON. = Rrdbridgc control area. 2 'I'hclt of a car/van/motor cycle, motor scootrr or mopcd. 3 Thcft from a carivanlmotor cyclc, motor srootcr or mopcd. 4 Criminal damagc to a motor vchiclc. 5 Burglary and attcmptcd burglary ot'a dwrlling and outbuildings. 6 Criminal damagr and attrmptcd damagc t o thc homc and dwclling site. 7 All housrhold of'cnccs including additional theft ollrnccs not shown in thc tablr. 8 All o n ~ n c c s against thr person including assault, thrft from thc person, robbrry, ollcnccr and threats falling within the survey's covrragr.

offence categories). Offences were included in the analysis i f they were reported as falling within the experimental period (twelve months before the interview) and if they were committed in the survey area. 'The distinction between single and series incidents made in the British Crime Survey analysis was used and the same rules for dealing with series incident were applied (see Hough and Mayhew 1985). Respondents who said that they had experienced multiple victimisations of a similar kind of offence were counted as being victimised up to a maximum of five times. Crimes were counted from the victim forms after being classified as offences as recommended in the Home Office counting rules.

'The proportion of households reporting one or more victimisations in the year M o r e interview is shown in Table 3.

Details of five household offences are shown along with the total numher of household offences (containing these and additional offence categories) and the total number of personal offences. T h e table shows that in both NW areas the prevalence (the number of households victimised one or more times) increased from the period before the launch of NW to the period after the launch. The prevalence of household victimisations also

248

increased in the displacement area. I t was only in the control area that the prevalence of household victimisation decreased.

The prevalence of reported victimisations against the person is much lower, but again there is no evidence of a reduction in either experimental area. The total number of offences against the person declined in the displacement area and increased slightly in the control.

Changes in victimisation rates of particular household offences are also shown in the table, but should be treated with caution because of the small numbers involved. It can be seen, however, that the broad trends described above arc reflected in most of the offence sub-categories. In Acton all household offence types increased from the pretest (before) to the post-test (after) surveys and in Wimbledon only burglary declined in prevalence. I t should be noted that the single percentage point reduction in burglary is based on small numbers and was not statistically significant.

Comparing the prevalence of victimisation between the two periods tells us little, however, about multiple victimisation and the total number of offences reported. Table 4 shows the mean offence rates per household reported in the two survey periods. The mean rate refers to the average rate of victimisation per household. This rate can be multipled by 100 to give a victimisation rate per 100 households or by 1,000 to give a rate per 1,000 households.

TABLE 4 Mean Offence Rate Per Household in the Pre-test and Post-test Periods in the Four

Survey Areas'

ACT.EXP. WIMB.EXP. W I M B . D I S P . RED.CON.

Pre Post Pre Post Pre Post Pre Post n = 306 309 353 400 323 332 306 330

Theft of vehicle .01 Theft from veh. .04 Crim. dam. to veh. .05 Burglary dwelling .08 Other crim. dam. .03

All h.hold offs.' .23

All pers. offs.' .03 (.02)

.03

.05

. I 1

.10

.05

.43 (.41)

.02

.04

.08

.06

.03

.28 (.27)

.06 (.04)

.01

.07

.16

.05

.09

.43 (.43)

.08 (.09)

.01

.07

.09

.09

. 1 1

.48 (.48)

.o 1

.06

.10

.07

.16

.01

.04

.06

.04

.04

.o 1

.02

.03

.02

.03

.48 (.48)

.13 (.13)

Notes: 1 Tablc headings and offence definitions are the same as shown in Table 3. 2 Means in brackets arc adjusted to take into account demographic and other differences between thc prc-test and post-test samplcs. 3 Personal offences have been weighted by a constant (number of personal offences reported/numbcr of residents in the arca aged I6 years or over) to standardise the responscs and to express personal offences as a rate pcr household.

249

TABLE 5 Mean Public Attitude and Behaviour Scale Scores f o r the Pre-test and Post-test

Periods in the Four Survey Areas

ACT.EXP.' W1MB.EXP. (n = 615) (n = 753)

Scale' Adj'd Fav./ Sig." Adj'd Fav./ Sig. Mean" Un Fav.' Mean Un Fav.

Fear of pers. vict." Pre Post

Pre Post

Pre Post

Pre Post

Pre Post

Pre Post

Pre Post

Pre Post

Pre Post

Fear ofh.hld. vict.'

Prob. ofpers. vict.H

Prob. ofh.hld. vict.!'

Satis. with area

Social cohesion

Home protection

Police evaluation

Policr contact

2.67 2.69

2.36 2.54

3.92 3.94

3.71 3.79

3.14 3.26

2.04 2.18

1.79 1.77

3.78 3.81

1.95 1.71

Fav.

Fav.

Fav.

Fav.

Fav.

Fav.

Unfav.

Fav.

Unfav.

2.49 2.6 1

2.29 2.37

3.92 3.86

3.70 3.68

2.98 3.10

2.17 2.24

I .82 1 .a9

3.62 3.58

2.54 I .76

*

*

*

*

Fav.

Fav.

Unfav.

Unfav.

Fav. *

Fav.

Fav. *

Unfav.

Unfav. *

Notes: I Table headings arc as shown in Table 3. 2 Scalrs comprise thc summation of responses (tom morc than one qurstion. A fullcr explanation ofttic construction of thc scales can be found in the text. 3 l'hc mran is adjusted to take into account demographic and other diffcrcnccs bctwccn thc prc-test and post-tcst samples. 4 Whcthcr the change lrorn the prc-tcst t o the post-test is in a favourablc or unfavourablc dircction. 5 Signilicant dillkrcnccs ( p < .05) arc marked with an asterisk (as determined by thc SPSSPC+ proccdurc, ANOVA). 6 Fcar of pcrsonal victimisation. 7 Fcar of household victimisation. 8 Pcrccivcd probability of personal victimisation. 9 Pcrccivrd ~robabi l i tv of household victimisation.

250

In both experimental areas the incidence (the total number of offences reported) of both household and personal victimisations increased from the pre-test to the post-test surveys. In the displacement area the rate of both types of offence remained constant and in the control area the rate went down for household offences and up for offences against the person.

It is possible, however, that these rates are misleading because the pre- test and post-test samples might have differed in ways which alone might have influenced the reported victimisation rates. For example, if households in higher income categories were more likely to be victimised than households in lower income categories, i t would be expected that a change in proportion of high and low income households would independently affect reported vicitimisation rates. In order to control for differences between the samples, it is necessary to use a statistical analysis which can simultaneously control for demographic and other differences between the samples. The figures in brackets at the bottom of Table 4 show the mean rates of victimisation after controlling for differences between the pre-test and post-test samples. Controlling for sample difference had little effect on the general finding that victimisation rates increased in the NW areas. In both areas the increase was statistically significant. Victimisation rates in the displacement area remained static from the ‘before’ to the ‘after’ surveys and decreased in the control area. There was no evidence, therefore, from these results of a NW effect on victimisation.

Crime prevention is a major aim of NW as conceived by the Metropolitan Police. Other aims include reduction in fear of crime and general improvements in the quality of life of residents. The impact of NW on attitudes and behaviour is shown in Table 5.

The table shows changes in the mean scores of scales used to represent concepts such as fear of crime and perceived probability of crime. The scales are constructed by adding together residents’ responses to a number of questions which relate to the underlying concept being investigated. Fear of household crime, for example, was constructed by summing together individual responses to questions concerning worry about burglary and worry about vandalism. The score presented in the table is the mean score for all respondents in each NW area and in each survey period. Low scores suggest that the mean is in an unfavourable direction and high scores show that it is in a favourable direction. A movement from the pre-test to the post-test from a low score to a high score represents an improvement in the behaviour or attitude being measured. In addition the mean scores used are adjusted to take into account differences between the pre and post-test samples.

Overall, there is a general movement from less favourable to more favourable scores on each of these factors. Not all of the changes are statistically significant, however, and not all are in a favourable direction. Fear of victimisation against the person declined in both NW areas, but the movement was not significant in either area. In other words, the change could have occurred by chance and cannot be counted as evidence of success. Fear of household victimisation declined in both areas and the decline was statistically significant in Acton, but not significant in

25 1

Wimbledon. Perceptions of the probability of victimisation moved in a favourable direction in Acton, but in an unfavourable direction in Wimbledon. In neither area was the change statistically significant. In both areas perceived satisfaction with the neighbourhood improved and both changes just reached statistical significance. Involvement with others, usually neighbours, in home protection worsened in Acton but improved significantly in Wimbledon. Evaluation of police performance improved in Acton and declined in Wimbledon. In neither area was the movement significant. Finally, contact with the police as measured by a single question concerning when the respondent last saw a police officer in their neighbourhood declined significantly in both areas.

Before summing up and attempting to interpret these findings it is worth including briefly some additional findings relating to public reporting of incidents to the police and the impact of NW on crime detection. I t was possible to ask all respondents who said that they or their household was victimised over the previous year whether the offence was reported to the police. I t is an aim of NW in London that there is closer co-operation between the public and the police and that the public report both crimes and suspicious incidents to them. It would be expected, therefore, that reporting rates would increase. In fact, reporting rates for household offences remained stable in Acton over the pre-test to post-test period and declined in Wimbledon. The number of telephone calls made to the local police from residents in the NW areas declined in both Acton and Wimbledon and the number of emergency (‘999’) calls increased slightly in Acton and declined in Wimbledon. The proportion of all calls concerning suspicious persons or incidents declined in both areas. Finally, there was no evidence that NW improved detection rates. Clear-up rates for the sub-divisions hosting the NW schemes went down slightly over thr experimental period.

Summary

The rrsults are most encouraging in terms of changes in the quality of life of residents in the NW areas. Fear of both household and personal crime move in a favourablr direction, although this was only statistically significant in relation to the Acton schemr. Improvements were also shown in residents’ satisfaction with their area in both Acton and Wimbledon and improvements in social cohesion in Acton and involve- ment with others in home protection in Wimbledon.

The findings of the research were less encouraging in terms of the impact of NW on crime in the areas studied. In addition, there was little evidence of success in terms of police clear-up rates, reporting rates or public reporting of suspicious incidents to the police. Finally, the reduction in the percentage of respondents reporting seeing a police officer in their area was both discouraging and surprising.

There is clearly a balance to be struck in assessing the favourable and unfavourable findings. It could be argued that evidence of any improve- ment in the areas should be treated as a NW success. I t is dificult to

252

argue, however, that the schemes showed evidence of an overall success when the primary aim of NW to reduce crime was not met.

Discussion

The general impression given by these results is that the changes shown were less promising than might have been hoped. How can this lack of a clearer and stronger NW effect be explained? There are a number of reasons why an evaluation might fail to find the desired effects. These have been summarised by Rosenbaum (1986) as ‘measurement failure’, ‘theory failure’ and ‘programme failure’.

Measurement failure means that the absence of a programme effect was the result of a poor evaluation design or method of statistical analysis which failed to detect what in reality was a programme effect. Theory failure means that the scheme was evaluated satisfactorily and was implemented satisfactorily but failed to show an effect because the theory behind the programme was flawed. In the case of NW it would have to be argued that the basic principles of NW were in error and no matter what effort was put into designing and implementing these programmes they would always fail to show evidence of success. Programme failure occurs when the evaluative design was correct and the theory of the programme was sound but the theoretical formulation was not adequately converted into practice. This could be because too little of the treatment ‘dosage’ was used or the wrong kind of ‘dosage’ was administered. In the case of NW it would have to be argued that the theoretical formulation was correct but it was not correctly operationalised and converted into practice in the schemes investigated.

Which of these explanations appears the most convincing? I t is hard to believe that the research design was so weak or badly conceived that it managed to mask what was in reality a much stronger NW success. I t is also hard to argue at this early stage in the history of NW that on the basis ofcurrent research evidence that the theory on which it is based is without substance (Rosenbaum 1986, 1987; Bennett forthcoming). The most convincing explanation of the limited success of the two NW schemes investigated is ‘programme failure’.

The issue of programme failure is double-edged. O n the one hand it might be that the specific programmes investigated were not good examples of NW in the MPD. In this case, it might be expected that other schemes that were designed more precisely to the guidelines would be more successful. O n the other hand, it might be that the design of NW in the M P D as expressed in the guidelines is not a good example of NW in general. There is some evidence to suggest that the latter case is the more correct.

I t is hard to believe that the schemes investigated were not good or at least average examples of NW in London. The co-ordinators of the scheme in Acton, for example, met regularly with the local police, held monthly co-ordinators’ meetings at which outside speakers were invited to attend, contributed actively to the creation of a NW newsletter and at the

253

end of the year held an anniversary street party to celebrate the scheme. I t was also clear that the local police were enthusiastic about the scheme and worked hard to make i t a success.

There are a number of reasons for believing that NW in the MPD might be short of the ideal. One important problem is the general lack of guidance in designing and implementing NW programmes. A principle element of NW in London is that divisional chief superintendents should be allowed flexibility in designing NW and fitting schemes to the needs and characteristics of local areas. In practice this has resulted in officers responsible for implementing NW doing no more than is stipulated in the guidelines and often a great deal less. In addition, the police have found it difficult to provide the resources to manage such an ambitious programme and in many areas are finding it difficult to deliver the goods. In the experimental areas, for example, it was shown that few residents had marked their property or had home security surveys. It was also shown that few residents had seen their local home beat officer recently. Uniformed patrol officers are called upon to carry out a wide range of duties which often means that they are taken off their beats to perform more reactive task.

The design of NW in London also might fall short of the theoretical formulation in that i t focusses perhaps too narrowly on the public becoming the eyes and ears of the police. In the United States, NW schemes have not focussed solely on the reporting of suspicious behaviour to the police as the principle mechanism by which NW is supposed to reduce crime. ‘The Seattle programme included the creation of signs of occupancy and more effective surveillance of each others’ properties (Cirel et al. 1977). It is surprising that the signs of occupancy has not been incorporated in the design of NW because it is a situational factor which has most frequently been reported in the research literature as capable of affecting offender decision making (Bennett and Wright 1984; Reppetto 1974; Walsh 1980)

The research reported concerns just two schemes in London and cannot be gcneralised to NW programmes in the rest of the country which might be of a different form and nature. Similarly, it cannot be assumed that other schemes in the MPD would produce similar findings. Nevertheless, the findings cannot be dismissed as applicable solely to the two schemes investigated. It must be considered why two very promising schemes failed to show a crime reduction effect. If the lack of crime reduction is a result of the design and implementation of the schemes then lessons can be learned from this one experiment

There is no reason to believe that the problems identified in the administration of NW schemes within the MPD are unique to this force. There is a danger that NW throughout the country is being implemented on the basis of uncertain theoretical principles and on speculative programme design. One of the important lessons that might be learned from this research is that the design and implementation of NW needs to be thought about carefully and honestly. In particular attention needs to be paid to the theory of NW and the crime preventing processes that the

254

programmes are presumed to instigate and t h e nature and design of the schemes tha t are supposed to evoke these processes.

Notes ’ Data were also collected on police recorded crime and on police time spent on

NW tasks. The results of this part of the analysis have been reported elsewhere (Bennett forthcoming).

References

Anderton, K . J. (1985) The Effectiveness of Home Watch Schemes in Cheshiw, Cheshire:

Assistant Commissioner ‘A’ Department ( 1983a) Force Instructions: Neighbourhood

Assistant Commissioner ‘A’ Department (198313) Neighbourhood Watch Schemes,

Bennett, T . H. (1987) A n Evaluation of Two Neighbourhood Watch Schemes in London

Bennett, T . H. (forthcoming) Evaluating Neighbourhood Watch, Aldershot: Gower. Bennett, T. H. and Wright, R. (1984) Burglars on Burglary: Prevention and the Oflender,

Aldershot: Gower. Cirel, P., Evans, P., McGillis, D. and Whitcomb, D. (1977) Community Crime

Prevention, Seattle, Washington: An Exemplary Project, Washington D.C.: L.E.A.A., U.S. Department of Justice, Government Printing Office.

Cook, T. D. and Campbell, D. T . (1979) Quasi-Experimentation: Design and Analysis Issues for Field Settings, Chicago: Rand McNally.

Hough, M. and Mayhew, P. (1983) The British Crime Survey (Home Office Research Study No. 76), London: H.M.S.O.

Hough, M. and Mayhew, P. (1985) Taking Account of Crime: Key Findings from the I984 British Crime S U ~ Z J ~ ~ (Home Office Research Study No. 85), London: H.M.S.O.

Judd, C. M. and Kenny, D. A. (1981) Estimating the Effects of Social Interventions, Cambridge: Cambridge University Press.

Reppetto, T. ( 1974) Residential Crime, Cambridge: Ballinger. Rosenbaum, D. P. (Ed.) (1986) Community Crime Prevention: Does it Work?, London:

Sage. Rosenbaum, D. P. (1987) ‘The theory and research behind neighbourhood watch:

Is it a sound fear and crime reduction strategy?’, Crime and Delinquency, 33 ( I ) ,

Russell, J. (undated) A Guide to Neighbourhood Watch Schemes, ‘A’ Department,

Walsh, D. (1980) Break-ins: Burglary from Private €louses, London: Constable. Washnis, G. J. (1976) Citizen Involvement in Crime Prevention. London: Lexington

Cheshire Constabulary.

Watch, London: Metropolitan Police.

Memorandum, London: Metropolitan Police.

(Report to the Home Office), Cambridge: Institute of Criminology.

103-34.

London: Metropolitan Police.

Books.

255